16270
Biochemistry 1996, 35, 16270-16281
An Essential Role for Water in an Enzyme Reaction Mechanism: The Crystal Structure of the Thymidylate Synthase Mutant E58Q†,‡ Carleton R. Sage, Earl E. Rutenber, Thomas J. Stout, and Robert M. Stroud* Department of Biochemistry and Biophysics, UniVersity of California, Box 0448, San Francisco, California 94143-0448 ReceiVed May 29, 1996; ReVised Manuscript ReceiVed October 11, 1996X
ABSTRACT: A water-mediated hydrogen bond network coordinated by glutamate 60(58) appears to play an important role in the thymidylate synthase (TS) reaction mechanism. We have addressed the role of glutamate 60(58) in the TS reaction by cocrystallizing the Escherichia coli TS mutant E60(58)Q with dUMP and the cofactor analog CB3717 and have determined the X-ray crystal structure to 2.5 Å resolution with a final R factor of 15.2% (Rfree ) 24.0%). Using difference Fourier analysis, we analyzed directly the changes that occur between the wild-type and mutant structures. The structure of the mutant enzyme suggests that E60(58) is not required to properly position the ligands in the active site and that the coordinated hydrogen bond network has been disrupted in the mutant, providing an atomic resolution explanation for the impairment of the TS reaction by the E60(58)Q mutant and confirming the proposal that E60(58) coordinates this conserved hydrogen bond network. The structure also provides insight into the role of specific waters in the active site which have been suggested to be important in the TS reaction. Finally, the structure shows a unique conformation for the cofactor analog, CB3717, which has implications for structure-based drug design and sheds light on the controversy surrounding the previously observed enzymatic nonidentity between the chemically identical monomers of the TS dimer.
Thymidylate synthase (TS;1 EC 2.1.1.45) catalyzes the reductive methylation of the substrate, deoxyuridine monophosphate (dUMP), using the cofactor, 5,10-methylenetetrahydrofolate (mTHF), to create the product, deoxythymidine monophosphate (dTMP), as a step in the sole de noVo pathway for the synthesis of dTMP. Since dTMP is required for cellular DNA synthesis, TS is the target for anti-cancer drugs such as the currently used chemotherapeutic agents 5-FU and ZD1694 (Jackman et al., 1991, 1993). The importance of TS as a chemotherapeutic target makes understanding the atomic details of the TS catalytic mechanism a critical part of the rational inhibitor design process. This understanding will also likely provide insight into the workings of the catalytic machinery of enzymes. The TS reaction mechanism has been heavily investigated, and the major steps of the reaction have been mapped [Figure 1; see Carreras and Santi (1995) and Stroud and Finer-Moore (1993) for review]. Most biochemical research on the TS mechanism has been performed using TS from Escherichia coli and Lactobacillus casei; therefore, L. casei residue numbers will be listed followed by the E. coli numbers in parentheses.2 The most important insight into the role of active site amino acids in the TS mechanism is provided by the high-resolution crystal structures of TS from various species complexed with substrates or substrate analogs [Hardy et al., 1987 (L. casei); Matthews et al., 1990 (E. coli); † Supported by NIH Grant CA41323 (R.M.S.), NIH Postdoctoral Fellowship AI09211 (C.R.S.), and an American Cancer Society Fellowship (T.J.S.). ‡ Crystallographic coordinates have been submitted to the Brookhaven Protein Data Bank under 1ZPR [E60(58)Q] and 1KCE (WT-TS). * Author to whom correspondence should be addressed. X Abstract published in AdVance ACS Abstracts, November 15, 1996. 1 Abbreviations: TS, thymidylate synthase; dUMP, 2′-deoxyuridine 5′-monophosphate; mTHF, 5,10-methylenetetrahydrofolate; dTMP, thymidine 5′-monophosphate; DHF, dihydrofolate; CB3717, 10-propargyl-5,8-dideazafolate.
S0006-2960(96)01269-X CCC: $12.00
Montfort et al., 1990 (E. coli); Perry et al., 1990 (E. coli); Fauman et al., 1994 (E. coli); Finer-Moore et al., 1994 (bacteriophage T4); Schiffer et al., 1995 (human)]. These atomic resolution views of the TS active site have redirected the use of site-directed mutagenesis to analysis of key residues in the active site. Although a number of highly conserved residues of TS have been extensively mutated, few substitutions abolish activity, presumably due to the “plastic” nature of the TS active site (Perry et al., 1990; Michaels et al., 1990; Climie et al., 1990, 1992). On the basis of crystallographic structural data (Montfort et al., 1990; Finer-Moore et al., 1990; Matthews et al., 1990a,b) and the absolute conservation (Carreras & Santi, 1995; Perry et al., 1990) of the glutamate at position 60 [E60(58)], it was proposed that E60(58) coordinates a chargestabilizing hydrogen bond network in the TS active site and thus plays a key role in the catalytic mechanism. E60(58) has been the subject of several site-directed mutagenesis studies assessing its role in the TS mechanism at the biochemical level. Zapf and co-workers (1993), analyzing the E60(58)Q mutant, proposed that the role of E60(58) in the TS reaction is either promotion of ring opening of mTHF upon ternary complex formation or stabilization of the cationic iminium intermediate of the cofactor (II, Figure 1). Two additional groups (Hardy et al., 1995; Huang & Santi, 1994) made the observation that mutations of E60(58) alter the rate-limiting step of the reaction so that dissolution of the covalent ternary complex between dUMP, mTHF, and TS became rate-limiting (IV, Figure 1) but differ in their interpretation of this result. Huang and Santi (1994) propose that E60(58) stabilizes a hydrogen bond network that 2 The amino acid sequence numbering used is that of the L. casei enzyme so as to be consistent with our previous publications. The E. coli sequence number is listed in parentheses. In addition, residues from the “second” monomer which enter into the discussion of the “first” monomer are indicated with a prime, e.g., R179(127)′.
© 1996 American Chemical Society
Structure of a Mutant Thymidylate Synthase
Biochemistry, Vol. 35, No. 50, 1996 16271
FIGURE 1: Proposed chemical mechanism and intermediates in the TS reaction. A box is drawn around intermediates which are proposed to be affected by E60(58)Q.
promotes proton transfer reactions at the substrate and cofactor and properly orients both ligands. Hardy et al. (1995) suggest that E60(58) acts Via electrostatic repulsion of O4 of dUMP to complete the methylene transfer reaction. To assess the role of E60(58) in the TS reaction mechanism at atomic resolution, we have determined the crystal structure of the E. coli TS mutant E60(58)Q cocrystallized with the substrate, dUMP, and the cofactor analog, CB3717, at 2.5 Å resolution. The conservative and isosteric glutamate to glutamine mutation at position 60(58) was chosen to allow observation of changes caused as a result of affecting only the functional group of the residue. The results support the notion that E60(58) coordinates a hydrogen bond network in which water plays an important role and illustrate the subtle changes in water position and hydrogen-bonding contacts that have drastic effects on the chemical reaction. In addition, the structure reveals an orientation for the ligands which sheds light on the nonequivalent enzymatic activity that has been observed between the two monomers of the dimer [reviewed in Carreras and Santi (1995)]. MATERIALS AND METHODS Site-Directed Mutagenesis A glutamate to glutamine point mutation at position 58 of E. coli TS was created by the method of Kunkel et al. (1987) using the synthetic DNA oligonucleotide 5′-CACAGCAGCTGATGGATG-3′ in the ThyA gene (Belfort et al., 1983) which had been cloned into Bluescript (Stratagene). Clones were first selected by screening the resulting plasmids for the loss of a NlaIII restriction site; positive clones were subsequently confirmed by DNA sequence analysis (UCSF, BRC Sequencing Facility).
Purification of Thymidylate Synthase Approximately 5 g of ThyA-strain χ2913recA (Climie et al., 1992) cells overexpressing E60(58)Q or wild-type TS (WT-TS) were obtained from 3 L of liquid LB culture grown for 18 h at 37 °C. TS was isolated as described by Maley and Maley (1988), and protein yields were similar for both E60(58)Q and wild-type TS [30 mg of TS/(g of cells)]. The purity of the isolated proteins was greater than 98% on the basis of Coomassie-stained SDS-PAGE gels. Enzymatic ActiVity Assays of TS The enzymatic activity of E60(58)Q and wild-type TS was assayed by monitoring the change in absorbance at 340 nm for the formation of H2-folate (Wahba & Friedkin, 1961). The specific activity of purified E60(58)Q and wild-type TS was similar to that observed previously (Zapf et al., 1993); however, it should be noted that Hardy et al. (1995) demonstrated that the change in absorbance at 340 nm does not reflect product formation for the E60(58)Q mutant but represents a stable chromophore. Product formation occurs at a rate 5-fold lower than that indicated by the spectrophotometric assay. Purified protein was stored as an 85% NH4(SO4)2 slurry at -20 °C. Crystallization For crystallization, purified E60(58)Q TS and WT-TS were dialyzed against 20 mM KPO4 (pH 7.5), 0.1 mM EDTA, and 1 mM DTT. E60(58)Q TS and WT-TS were cocrystallized with the folate analog CB3717 and dUMP as described previously (Montfort et al., 1990). Large, single crystals grew over a 3 week period, and 2 days before data
16272 Biochemistry, Vol. 35, No. 50, 1996
Sage et al.
collection, β-mercaptoethanol was added to the wells to a final concentration of 1 mM to ensure the protein was under reducing conditions. For E60(58)Q TS, two data sets were collected from separate hexagonal rod crystals 0.25 mm in diameter and 0.40 mm long. For wild-type TS, a single data set was collected from a hexagonal rod 0.3 mm in diameter and 0.5 mm long.
positional and B factor refinement in X-PLOR (Bru¨nger, 1992) followed by manual inspection of electron density maps using the program CHAIN (Sack, 1988). The criteria for water addition were difference density greater than 3σ in Fo - Fc maps and located within 3.5 Å of a hydrogen bond mate. The final R factor is 15.3% to 2.5 Å. The Rfree is 24.0%. The E60(58)Q structure has been submitted to the Brookhaven Protein Data Bank as entry 1ZPR. WT-TS. In order to directly compare mutant and wildtype diffraction data, it is important that the diffraction data be collected on a similar instrument and scaled and refined using equivalent software. Since the original TS-dUMPCB3717 structure (Montfort et al., 1990) was solved using data from a multiwire detector and was not scaled and refined using the software used here, we recollected data from TSdUMP-CB3717 crystals and re-solved the structure by isomorphous molecular replacement using a dimer of the E. coli ternary complex (Montfort et al., 1990), including waters and ligands, as the starting model. Successive rounds of positional and B factor refinement in X-PLOR (Bru¨nger, 1992) followed by manual inspection of electron density maps using the program CHAIN (Sack, 1988) resulted in a solution which was nearly identical to the starting model, but has better geometry, and includes the N-terminal modification of E. coli TS (Fauman et al., 1994). The final R factor is 18.9% to 1.95 Å. The Rfree is 23.7%. The resolved structure has been submitted to the Brookhaven Protein Data Bank as entry 1KCE.
Data Collection
RESULTS
X-ray diffraction data were collected on an R-Axis IIc imaging plate with a Rigaku RU-200 rotating anode generator operating at 15 kW (50 mA and 300 kV) fitted with a Cu anode (λ ) 1.5418 Å). The crystal-to-detector distance was 90 mm. Exposures of 20 min/(deg of oscillation range) were used throughout the data collection, and 90 [E60(58)Q] or 75 (WT-TS) degrees of data were recorded for each crystal. The diffraction data to 2.3 Å (E60(58)Q) or to 1.95 Å (WTTS) resolution were indexed, integrated, scaled, and merged in the hexagonal space group P63 with the following unit cell dimensions using the HKL software package (Otwinowski, 1990): a ) b ) 127.2 Å and c ) 68.09 Å. A summary of the data-processing statistics is presented in Table 1. For the E60(58)Q crystals, a total of 200 624 observations were integrated, scaled, and merged, yielding 26 346 unique reflections between 50 and 2.3 Å [Rmerge(I) ) 8.5% with average redundancy of 7.6]. For the WT-TS crystal, a total of 221 970 observations were integrated, scaled, and merged, yielding 37 807 unique reflections between 50 and 1.95 Å [Rsymm(I) ) 9.7% with average redundancy of 5.8].
In order to address the role of E60(58) in the TS reaction at atomic resolution, the TS mutant E60(58)Q was created using oligonucleotide-directed mutagenesis. TS isolated from a ThyA- strain overexpressing E60(58)Q TS showed activity similar to that previously observed (Zapf et al., 1993). Purified E60(58)Q TS was cocrystallized with the substrate, dUMP, and a quinazoline-based antifolate inhibitor, CB3717. Diffraction data were collected, and the structure was solved by molecular replacement. The E60(58)Q model was refined at 2.5 Å to a final R factor of 15.2% (Rfree ) 24.0%). The refined model has good geometry (Table 1), no Ramachandran outliers, and a 2Fo - Fc electron density map contoured at 1.0σ is well-defined for most of the molecule. As previously noted (Montfort et al., 1990), this crystal form contains one TS dimer in the asymmetric unit, and each monomer in the dimer is related by an approximate noncrystallographic two-fold (κ ) 178.5°). In the TS dimer, each monomer contributes two phosphate binding arginine residues to the active site of the other monomer. To describe their participation from the second monomer, these residues are referred to as R178(126)′ and R179(127)′. In this study, the orientations of the ligands in the two active sites differ significantly; therefore, each monomer will be considered independently and compared to the corresponding monomer in the re-refined structure of wild-type E. coli TS in complex with dUMP and CB3717.
Table 1: Data Collection and Refinementa space group unit cell dimensions Rmerge 〈I/σ(I)〉 completeness (50-2.5 Å) observations unique reflections refinement resolution R factor Rfree rms deviation from ideal geometry bond lengths bond angles dihedral angles improper angles Ramachandran outliers
E58Q
WT-TS
P63 a ) b ) 127.2 Å, c ) 68.2 Å 8.5% (26.5) 7.4 (2.8) 93.8% (84.6) 200 624 26 346 7.0-2.5 Å 15.2% (20.9) 24.0% (29.6)
P63 a ) b ) 126.8 Å, c ) 67.8 Å 9.7% (26.1) 7.3 (2.4) 90.0% (69.3) 221 970 37 807 7.0-1.95 Å 18.9% (28.2) 23.7% (34.6)
0.014 Å 3.20° 25.81° 1.37° none
0.008 Å 1.61° 25.54° 1.30° none
a Data shown in parentheses are those for the highest-resolution bin determined.
Structure Solution E60(58)Q. The E60(58)Q TS structure was solved by molecular replacement using two monomers of the E. coli TS product ternary complex with ligands and waters removed (Fauman et al., 1994) superimposed onto the E. coli ternary complex (Montfort et al., 1990) as the search model. Rigidbody refinement using the program X-PLOR (Bru¨nger, 1992) gave an initial R factor of 31%. Ligands and waters were located from Fo - Fc difference maps (Chambers & Stroud, 1977) during successive rounds of simulated annealing and
E60(58)Q Compared to Wild-Type E. coli TS Monomer 1 Monomer 1 [chain A in PDB entry 1KCE for wild type; chain A in PDB entry 1ZPR for E60(58)Q] of the E60(58)Q TS structure was compared to monomer 1 of the
Structure of a Mutant Thymidylate Synthase
Biochemistry, Vol. 35, No. 50, 1996 16273
FIGURE 2: Stereo representation of the superimposition of the active sites of monomer 1 of wild-type E. coli TS (black) and E60(58)Q TS (white). This figure was produced using the program Molscript (Kraulis, 1991).
FIGURE 3: Stereo representation of a refined Fo - Fc omit map of the active site of monomer 1 contoured at 3.0 σ. This map was calculated after C198(146), dUMP, CB3717, water1, water229, and waterC7 were omitted from the PDB file and the coordinates subjected to 10 cycles of positional refinement. This figure was produced using the program Molscript (Kraulis, 1991).
re-refined structure of wild-type E. coli TS cocrystallized with the substrate dUMP and the cofactor analog CB3717 using the program GEM (Fauman et al., 1994). After superimposition of monomer 1 from the two structures, the rms deviation of CR atoms is 0.286 Å, and the rms deviation of all protein atoms is 0.753 Å. The rms deviation for all dUMP atoms in monomer 1 is 0.461 Å, and the rms deviation for all atoms in CB3717 is 0.689 Å. In this monomer, the average B factor for the wild-type structure is 17.97 Å2 and for E60(58)Q is 18.59 Å2. Comparison of ActiVe Site 1. As shown in Figure 2, the overall arrangement of the active site residues in E60(58)Q TS is almost identical to that of the wild-type structure. As observed in the wild-type ternary complex structure (Mont-
fort et al., 1990), a covalent bond is formed between the Sγ of C198(146) and C6 of dUMP (Figure 3). The main differences in residue positions between the two structures are small. The side chains of W82(80), Y146(94), F228(176), and S232(180) exhibit the most dramatic movements (Figure 2). The side chain of W82(80) rotates by 10° around the Cβ-Cγ bond out of the active site as compared to that of the wild-type structure. The side chain of Y146(94) moves out of the active site by 0.36 Å. The side chain of F228(176) translates out of the active site by 0.8 Å. The S232(180)Oγ flips to a second rotamer 140° away. Major differences occur in the position of water molecules3 whose location is strongly conserved in several other TS structures. Water229a is no longer constrained in the structure and not
16274 Biochemistry, Vol. 35, No. 50, 1996
Sage et al.
Table 2: Comparison of Contacts in Active Site 1 distance (Å) interaction phosphate R218(166)NH1-PO4 R218(166)NH2-PO4 R218(166)NH2-PO4 R178(126)′N-PO4 R178(126)′NH1-PO4 R179(127)′N-PO4 R179(127)′NH1-PO4 R23(21)N-PO4 S219(167)Oγ-PO4 ribose R23(21)NH1-OR5 R178(126)′NH1-OR5 Y261(209)Oζ-OR3 H259(207)N-OR3 S219(167)Oγ - OR3 uridine N229(177)Oδ-N3 N229(177)Nδ-O4 water1-O4 D221(169)N-O2 Q217(165)N2-O2 Y146(94)Oη-C5 waterC7-C5 pterin ring G225(173)N-OA4 A315(263)O-NA2 D221(169)Oδ-N3 water315-CB3717 N1 water221-CB3717 NA2 E60(58)/Q water229-E60(58)QO/N W82(80)N-E60(58)QO water232-E60(58)QO/N water1-E60(58)QO other important interactions N229(177)Nδ-water229 water1-H199(147)N waterC7-Y146(94)Oη A315(263)O-water315 W85(83)N-I316(264)OT waterC7-A196(144)O water221-A315(263)N water315-I316(264)OT water229a-water229 water229a-S232(180)Oγ water232-I57(55)O water232-water229a water229-H199(147)N2 water229-water1 water232-S232(180)Oγ a
wild type
E60(58)Q
2.66 2.71 3.35 3.04 3.32 3.02 2.77 2.87 2.84
2.83 2.75 3.38 3.03 3.21 3.44 2.98 3.99 2.97
3.16 3.49 2.77 2.84 4.20
3.11 2.94 3.03 2.83 4.34
2.98 3.14 2.89 3.15 3.13 3.86 3.91
2.95 3.25 3.25 3.05 3.17 3.87 3.32
3.02 2.89 2.70 2.89 3.11
3.01 2.83 2.81 3.29 3.10
2.95 2.85 2.74 2.89
3.58 2.95 2.90 2.78
2.91 2.83 2.67 2.89 3.17 2.90 3.17 3.00 3.01 2.79 2.83 2.69 4.55 3.67 2.69
2.74 2.90 2.82 3.75 2.93 3.44 2.82 2.84 NPa NP 2.83 NP 3.02 2.98 2.62
NP ) not present.
defined by crystallographic means. Water229 is moved 1.75 Å, water1 0.34 Å, and waterC7 0.75 Å. A comparison of the intramolecular contacts in the active site is shown in Table 2; several important differences among these contacts are observed between the two structures in the active site. The distance between water229 and either the N [E60(58)Q] or the O (WT) of residue 60 increases from 2.90 Å in the wild-type to 3.58 Å in the mutant structure. 3 Waters are numbered according to the number or name of the residue to which it is closest with the exception of water1 and waterC7 which were described in Fauman et al. (1994). If more than one water is close to a single residue, the waters are further designated with letters. For example, there are two waters around N229; the first is designated water229 and the second water229a.
The distance from waterC7 to C5 of dUMP decreases from 3.91 Å in the wild-type structure to 3.32 Å in the mutant structure. In addition, the contacts of waterC7 with the protein have changed. In the wild-type structure, waterC7 makes a 2.67 Å hydrogen bond with Y146(94) and a 2.90 Å hydrogen bond with the main chain oxygen of A196(144), but in the E60(58)Q structure, these hydrogen bonds have been weakened to a 2.82 Å hydrogen bond with Y146(94) and a 3.44 Å contact with the main chain oxygen of A196(144). Most importantly, the distance between water1 and O4 of dUMP increases from 2.89 Å in the wild-type to 3.25 Å in the E60(58)Q structure, weakening the hydrogen bond between water1 and dUMP and creating a new hydrogen bond between water1 and water229. A relative comparison of B factors provides insight into the relative disorder of one structure Versus another (Table 3). The overall B factors for the wild-type and mutant structures are similar: 17.97 Å2 for the wild-type and 18.59 Å2 for the mutant. In order to make a relative comparison of the thermal motion between the two active sites and to avoid the problems with a direct comparison of B factors, the B factors of the atoms in the active site were normalized to the B factors of the J helix [residues 225(173) to 245(193)], the innermost part of TS, to determine a relative disorder for each active site. Comparison of the normalized B factors for atoms in the active sites reveals that the mutant active site (average B ) 17.38 Å2; relative disorder ) 1.17) is more disordered than the wild-type (average B ) 12.79 Å2, relative disorder ) 0.93) active site. This 25% increase in relative disorder may also contribute to the reduction of catalytic activity of the mutant enzyme. E60(58)Q Disrupts the ExtensiVe Hydrogen-Bonding Network Necessary for Proper TS Function. In a structure solved to 2.5 Å, hydrogen atoms are not discretely observed; however, with careful analysis, the hydrogen bond contacts can be inferred. An important means of elucidating the differences between two homologous structures is the calculation of an unbiased difference map. To directly compare the atomic differences between the wild-type and E60(58)Q structures and aid in the deduction of the hydrogen bond networks, newly collected diffraction data from wildtype TS cocrystallized with dUMP and CB3717 were scaled with diffraction data from E60(58)Q TS, also cocrystallized with dUMP and CB3717, and an Fo[E60(58)Q] - Fo(wild type) difference map was calculated with phases from the rerefined wild-type TS model using XtalView (McRee, 1993). Shown in Figure 4 is the difference density around residue 60(58) contoured at 3.5σ along with the overlapped structures. Several important observations are made from this map. The side chain oxygen of serine 232(180) has rotated to a new rotamer 140° away from the wild-type, and difference density for the elemental change from O to N at the position of the E60(58)Q mutation is visible. There is WT difference density for both waters around N229(177) and E60(58)Q difference density for the single water which connects N229(177) and E60(58). As expected, the manner in which Q60(58) interacts with the hydrogen bond network is different than that in the wild-type structure. It is striking that, although the manner in which water232 interacts with the protein is now different (Figure 5), its position is nearly identical with that of the wild-type (Figure 2). Figure 5A shows the inferred wild-type hydrogen bond network interacting with dUMP. In this hydrogen bond
Structure of a Mutant Thymidylate Synthase
Biochemistry, Vol. 35, No. 50, 1996 16275
Table 3: Average B factors (Å2) for Specific Groups of Atoms WT-TS
E60(58)Q
group
monomer 1
monomer 2
monomer 1
monomer 2
all CR’s all protein atoms all atoms of the J helix all atoms of dUMP all atoms of CB3717 all protein atoms
